How AI Actually Sees You: The Two Doors to AI Visibility

AI can only reach you through two doors. Door number one is memory, which means AI knows you by heart and can talk about you without searching the internet. Door number two is live search, which is when AI finds you while someone is searching and it scans the web. You need to build both, and you need to stop assuming there is only one way in.

AI visibility: being findable through both doors, live search and model memory, so AI can cite you, recommend you and eventually name you without having to look you up.

Memory is the most durable form of AI visibility and the highest thing you can achieve, but it takes time, because you have to earn it. To get inside Claude’s memory, ChatGPT’s memory or Perplexity’s memory, your name has to appear on the web a lot of times over a long period. Those models are trained and updated on a regular basis, and when your name comes up over and over again, at a certain point it gets included.

Live search is faster to influence, but it is more fragile, because you are only visible while your page is ranking and while the user has search turned on. Most users do have search on these days, so that part is not something to worry about, but you still have to show up regularly.

So you start at door number two, live search, and the more consistently you show up online, the more likely it becomes that you end up inside AI’s knowledge. Most people put all their effort into the search door and never think about the memory door, which is backwards. AI either remembers you or it looks you up. Build both.

How You Get Cited by AI

Getting cited means AI reaches your website and uses the content you created to answer a question someone is asking inside a large language model. For that to happen, your content has to answer the specific questions people are actually asking.

This is the big difference between SEO and GEO. With search engine optimisation, we found a keyword with a lot of traffic and volume, wrote the article around it, added the keyword as often as we could without stuffing it, and hoped it would rank. With generative engine optimisation, we write for user intent. You have to think very clearly about what your ideal client is actually asking, create that content, and publish it on your website.

Besides creating the content, your website also needs to be AI readable, and for that you have to add schema markup. There are multiple types of schema markup. A schema markup is a snippet of code you add to your site, and you need it because AI reads a website differently than humans do. We go to a website and read the content. Machines do not read the text, they read the code and the data you add to the site.

Big media coverage helps here too. Your website is your hub, and we will come back to that, but if you are published in a large publication and you answer the questions your ideal clients are asking in that article, the chances are high that AI will cite you, because it is a high authority website and you gave the answer.

The catch is that this visibility is real time. It can appear the week you publish and it can disappear the moment something else ranks above you, which is why you have to put out content regularly on your website and in the media. Treat citations as your fast lever, not your foundation. They are amazing, because your content gets cited and people can click through and discover you, but they move up and down with search results every single day. Citations are rented visibility. They work fast, and they last only as long as you keep ranking.

How You Get Recommended by AI

The second thing is getting recommended, which is AI naming you as the expert. When someone asks AI for the best expert in your niche, the model answers from a sense of who the web agrees is credible. You want to be that name.

So how does the web agree who is credible? Through third-party signals, which is what other people say about you rather than what you say about yourself. If you say you are great, that is advertising. If your neighbour says you are great, that is PR.

What we have to do is get mentions and links from other websites, from podcast appearances, from media features and media appearances, from being published, and from a Google Knowledge Panel. We need to tell AI that the wider web treats you as authoritative. Because AI cannot read content the way humans do, it needs trust signals, and a trust signal is always a third-party trust signal. Advertising is not trustworthy, because you can say anything about yourself. When somebody else says it, it becomes more trustworthy.

The more trustworthy the source, the heavier the trust signal. Getting mentioned or featured in your local newspaper online is great, and it is a third-party endorsement. But if you get featured in Forbes, you can see the difference. Forbes has higher authority, so the trust signal is higher and more valuable to AI.

Your website and your schema markup are the hub of all of this, together with consistent naming, so AI can connect every endorsement back to your identity. I know there are a lot of people out there saying you do not need a website, and I agree, you do not need one if all you want to do is make sales. But if you want to build a personal brand, a business, something that is AI future-proof, you need a website. Period.

The other piece is entity consistency, which means you call yourself the same thing everywhere. Same name, same title, same description of what you do, the same elevator pitch, everywhere. My middle name is Elizabeth, but it is never Simone Elizabeth Sauter in one place and Simone Sauter in another and Simone S. somewhere else. It is always Simone Sauter. And it is always PR and AIO strategist. Not PR expert, not podcast expert, not AI expert. That consistency is what lets AI connect the dots.

A recommendation can come from the model’s memory and from live search, which is why the two overlap. By investing in speaking on podcasts, in being in the media, in getting a Google Knowledge Panel and in getting mentions and backlinks of all kinds, you invest in both doors at once. You get recommended when the whole web agrees you are the expert, not just your own site.

How You Get Remembered by AI

The third thing is being remembered, which means you live inside the model. AI knows you from its training data rather than from a live search, so it can talk about you with search turned off, without looking you up.

This is the strongest position, because it does not depend on rankings, on freshness, or on whether the user has search enabled. Your name and your brand live inside the model. And as I said at the start, you earn it slowly, through consistency. Your name has to show up consistently across the web, on multiple pages, with entity consistency holding it together.

The most important part of living inside the memory is having a Google Knowledge Panel. You get there through a lot of interconnected work: feeding the Google Knowledge Graph, looking at your Wikidata, building the backlinks. Everything I teach my clients connects, and there is a structure to it, and that structure exists to get you visible through both doors.

For memory, the Knowledge Panel matters enormously, because it is not only a trust signal. It is proof that you have become a recognised entity, which is exactly what gets you into the model. A recognised entity means Google and the Google Knowledge Graph understand that you are an authority, and AI relies heavily on the Google Knowledge Graph. That is what we need to train.

Because the models are retrained over time, the work you do now to get mentioned and quoted compounds quietly into being known by future versions. Citations fade and recommendations shift, but being baked into the model is visibility that you own. This is why it is so crucial to invest in your AI visibility now.

Where to Show Up So It Counts

I get asked constantly where to show up, so here is the order.

First and most important is the Google Knowledge Panel. It is the highest level you can reach in AI visibility, and in brand building generally. Media comes next, and with media, the bigger and more known and more trustworthy the publication, the better. Then podcasts. With a podcast you get a backlink, and you create long-form natural language content that can be cited. Media can be cited too, and it gives you the backlinks. That is why media and podcast appearances are so crucial: you create content that can be cited, you create backlinks so you can get recommended, and you build the online structure you need in order to be remembered.

Then there are platforms you can add, and I am not saying you should add all of them. My strongest recommendation is LinkedIn. It is the second most cited website out there, so I publish content on my website and republish it on LinkedIn. But there is another reason. When we pitch media and podcasts, the first thing every editor and journalist does is check whether you are on LinkedIn and look at your profile. You do not have to publish there every single day, but I strongly recommend a professional profile that reflects who you are and what you do.

YouTube is another great platform, because AI can read the content, you create content that can be cited, and it is one of the biggest search engines out there. I also get asked about Substack a lot. Substack is a great place to be, and it gives you a lot of the functions you need, from schema markup to internal linking, to get cited.

Still, my recommendation is always the same. Publish the content on your website first, then republish it on Substack, on LinkedIn, on Medium, wherever you want. Build your website and treat it as your hub, because everything comes together there. Your credibility, but also the experience of a potential client landing on your site, seeing your media appearances and podcast appearances, seeing what you are about, reading your blog and finding that you answer all their questions. That builds instant trust.

A lot of people are on Skool as well, and I am on Skool myself. There you have the option to set your community to public, which depends on your niche, but if it is public then large language models can read it, so good content there is readable to AI too.

What is not relevant to your AI visibility is no-name blogs and social media. It simply is not, because what we say on social media is basically advertising. It is us saying we are amazing, talking about our products, talking about what we do. That is not interesting to AI, and it is not trustworthy to AI.

Frequently Asked Questions

Do Not Try to Be Everywhere

The takeaway is this. Do not try to be everywhere. Win the few sources AI actually trusts, start with the Google Knowledge Panel, and let those carry you. To get a Knowledge Panel you have to put all the dots together, which is exactly why everything I teach is interconnected.

If you want to learn more, join the AI Visibility Lab, where we come together and go deeper on all of this.